This paper reports on a series of satellite trials using the Australian Optus B3 spacecraft. A Eureka 147 signal was successfully transmitted from an L-Band communications payload aboard the satellite and the reception of CD quality audio demonstrated under both mobile and fixed site conditions. Also included in the trials were an investigation of the effect of transponder operating point on intermodulation levels, using the wide-band Eureka signal, and a study of the likely link margin requirements for an operational L-Band digital sound broadcast service. The latter involved signal level measurements in a variety of environments and at three separate elevations, representative of the mid-to-lower range of geostationary angles in Australia.

1. Introduction
2. Optus L-Band communications payload
3. Equipment configuration
3.1 Up-link equipment
3.2 Receiver equipment
4. L-Band transponder operating point
5. Satellite link budget
6. Audio monitoring programme
7. Transponder linearity
8. Measurement of signal fade distributions at L-Band
8.1 Measurement system
8.2 Measurement procedure
8.3 Data processing
8.4 Presentation of results
9. Discussion of narrow-band fade results
9.1 Open areas
9.2 Rural, suburban and dense suburban areas
9.3 Wooded areas
9.4 Urban, dense urban and highrise areas
10. Shadowing and multipath contribution to signal variability
11. Summary & Conclusions
12. Acknowledgements

ANNEX 1 Photographs of equipment configuration
ANNEX 2 Cumulative fade distribution curves as a function of elevation angle and clutter type
ANNEX 3 Narrow-band cumulative fade data (50, 10, 1%) & standard deviation values for each sector (Canberra, Adelaide & Perth)

This paper reports on satellite broadcast trials using an L-Band communications payload aboard the Australian Optus B3 spacecraft. The experimental trials were undertaken by the Communications Laboratory as part of an on-going programme of studies into the Eureka 147 digital sound broadcast system.

The main objectives of these trials were to:

1. demonstrate the operation of the Eureka 147 system over an operational satellite L-Band communications payload;
2. undertake a qualitative assessment of the performance of the Eureka system in a variety of reception environments;
3. collect data on the fade statistics of the L-band mobile satellite channel for a range of reception environments and elevation angles.

The satellite trials were implemented in two phases over the period June 1995 to February 1996. The first phase of measurements was undertaken in Canberra and the surrounding regional area during June/July 1995. This was followed by a second series of measurements in Adelaide (December 1995) and Perth (January 1996).

The Optus B3 satellite is based on a Hughes 601 space platform and was launched in late 1994. At time of commencement of the trials the satellite was in a slightly inclined orbit at 152° E, in preparation for entry into service in late August 1995, at a location of 156° E.

The L-Band payload aboard the spacecraft operates in the 1545 - 1559 MHz frequency range and comprises eight solid state power amplifiers (SSPA), which combine to provide a total output power of 150 Watts. The package is designed for the MobileSat communications service and utilises L-Band transmission and reception to mobile stations, and Ku-Band communication between the satellite and ground base station. The service foot print extends over continental Australia and, for a fully loaded MobileSat service, provides an EIRP of 49 dBW at beam centre [1]. The corresponding signal level in Adelaide is 48 dBW, and somewhat less in Canberra and Perth (46.8 dBW and 46.1 dBW respectively).

3.1 Up-link equipment

Figure 1-1 shows a schematic of the encoding and up-link equipment. The digital broadcast was based on third generation Eureka 147 hardware (Philips PDS). The equipment is capable of generating a DSB signal at an intermediate frequency of 36 MHz using any one of the three transmission modes identified for the Eureka system. The encoded signal has a fixed frame format allowing selection of one of nine different applications, corresponding to audio and data channels of varying bit rate and protection level. For the trials transmission Mode II was used. The audio programme from the CD player was 'compressed' using the DSB system's internal MUSICAM encoder at a bit rate of 256 kbits/sec (independent stereo mode) and channel coded using an average rate of 6/10. The output of the encoder was then 'shifted' to an intermediate frequency of 70 MHz and up-converted to Ku-Band for up-linking to the B3 spacecraft.

During the first phase of the trials, the encoded signal was up-linked from the Optus Belrose (Sydney) earth station using a 4.5 m tracking antenna. For the second phase the signal was up-linked from a 4.6 m antenna, located at the Communications Laboratory (Canberra).

3.2 Receiver equipment

The corresponding hardware configuration for monitoring of the received satellite signal is shown in Figure 1-2. The down-link frequency of 1552.5 MHz is only 4-7% above the nominal DSB L-Band allocation (1452-1492 MHz), so it is reasonable to expect results to correlate well with operation in this band. For mobile reception a bifilar helix rod antenna, designed for the L-Band mobile telephony service was used. This antenna has an omnidirectional pattern in the horizontal plane but an inclined pattern in the vertical axis which is adjustable to match the elevation angle of the satellite. The nominal peak gain of the antenna is 8.5 dBic. For fixed monitoring a directional panel antenna with gain of 14 dBic was used. The receiver front end consisted of two cascaded low noise amplifiers, followed by a mixer stage, and a third generation Eureka receiver for decoding of the signal. The recovered audio signal was monitored using both headphones and high quality loud speakers.

The gain transfer curves of Travelling Wave Tube Amplifiers (TWTA) show a characteristic flattening, and a so called ‘saturated’ power output can be clearly defined. In contrast SSPAs have a less clearly defined output limiting characteristic. As the input level is increased, SSPAs exhibit increasing gain compression, and dissipate greater amounts of heat, eventually failing due to thermal breakdown. With no inherent protection on the amplifiers aboard the spacecraft, up-link ‘overdrive’ can cause thermal failure.

With the Optus B3 L-band payload the single carrier saturated output (or 0 dB OBO) of the non-linearised transponder is set arbitrarily to the 2 dB compression point. However, the SSPAs are not rated for continuous operation at this level. The maximum safe single carrier output power level is 2.2 dB less (i.e 2.2 dB OBO), and it is this level which is also deemed to be the maximum continuous operating point for multi-carrier operation.

For the trials a further 1 dB of input back-off (IBO) was applied, resulting in an additional 1.7 dB OBO and a corresponding down-link EIRP of 46.3 dBW (Adelaide), 45.1 dBW (Canberra) and 44.4 dBW (Perth).

Table 1-1 provides a summary of the satellite down-link budget based on parameter values for each of the three locations. The link budget does not include implementation allowances (such as for hardware, interference and fading), normally expected of an operational service. Because of the relatively low transponder power it was necessary to improve the receiver figure-of-merit (G/T) by using a relatively high gain antenna and by incorporating a two stage low noise amplifier at the

front-end of the third generation Eureka receiver. These changes resulted in a 7 dB improvement over the nominal G/T figure anticipated for an operational service [2].

The link budget calculations indicate that for mobile reception the receiver system is 0.1 to 2.3 dB above receiver threshold, depending on location. In the case of fixed reception, using the higher gain panel antenna, the corresponding margin ranges from 3.7 to 5.9 dB. Based on these preliminary calculations it was anticipated that mobile reception would be marginal, particularly in Perth and Canberra, while fixed reception could be expected to be satisfactory.

Initial attempts to up-link the Eureka 147 signal to the OPTUS B3 satellite revealed transient variations in the spacecraft power bus. These variations were found to occur once every 24 ms, and coincided with the null symbol period for Mode II during which no RF signal is transmitted. Because of these observed anomalies and concerns regarding the possible effect on both the L-Band SSPAs and the rest of the payload, testing was temporarily discontinued and the spacecraft manufacturer consulted. The subsequent advice received indicated that there was no risk to either the SSPAs or the payload. The spacecraft manufacturer also confirmed that operation could be permitted up to the multi-carrier operating level (i.e 2.2 dB OBO).

Audio monitoring of the satellite DSB signal was undertaken, initially, in Canberra. Using the fixed panel antenna reception was consistent with no impairment of the decoded audio signal. For mobile reception, test routes included sections of flat open and suburban areas, as well as the main business district which is characterised by multi-story buildings (typically 6-7 floors). In open and suburban areas the reception quality was judged as consistently good. Because of the relatively high elevation of the satellite (51º ), blockage of the signal in both these environments was infrequent, and then, usually due to overhanging trees along the roadside. As expected, the receiver suffered more frequently from signal blockage in the built up business district. Depending on the relative orientation of the satellite and buildings the receiver would generally mute. It was interesting to observe that in this environment there is little evidence of audio signal degradation before the onset of muting. The presence of large building obstructions, together with the relatively low vehicle speed, results in abrupt and extended blockage of the signal, ensuring that the transition from perfect reception to complete muting occurs very ‘cleanly’. Similarly, once clear of building obstructions the presence of a relatively constant signal level enables the receiver to recover quickly, and in an equally clean manner.

The audio failure characteristic was more disconcerting in other mobile reception situations - most notably where the satellite path was obstructed by closely spaced features such as trees, or by foliage of varying density. Here, the affect is similar to that observed with terrestrial operation (although the former reflects mostly shadowing rather than multipath impairment), where rapid fluctuations in received signal level below and above threshold can result in a more objectionable failure characteristic.

Given the encouraging results from the Canberra trials, an extensive audio monitoring programme was undertaken along some 4,400 km of road distance, extending from Canberra to Adelaide, and on to Perth. The results of this exercise were equally as promising. In rural areas between Canberra and Adelaide, mobile reception was generally good. Although audio impairment was evident in the presence of all, but the most minor obstructions (i.e roadside light or power poles, overhanging power lines, or individual lightly foliaged trees), this was expected given the low link margin. Even so, the relatively high satellite elevation angle and mostly open reception conditions ensured such occurrences were infrequent.

The most consistent reception was along the national highway linking Adelaide and Eucla (a small roadhouse community just within the West Australian border). The 1350 km section of highway traverses mainly flat terrain with either low shrub or scattered tree cover, the latter set well back from the road. Reception along this section was virtually continuous (i.e approaching 100% of locations) and coincided with maximum signal level for the entire route - estimated at 2.7 dB above receiver threshold. The (very) occasional audio ‘breaks’, were due to antenna misalignment, caused by flexing of the 1.0 m MobileSat rod antenna at highway speeds (which at times exceeded 120 km/hr).

Travelling from Eucla, along the Eyre Highway, the open shrubland gave way to areas of woodland near the historic mining community of Norseman (200 km south of Kalgoorlie), resulting in more frequent signal outages due to road-side shadowing. This, and the remaining leg of the journey to Perth was accompanied by a progressive reduction in received signal level, associated with the satellite foot-print. Despite this, satisfactory reception was still possible along the mainly open section of the Eastern Highway between Kalgoorlie and Tammin (approximately 180 km east of Perth).

The results of this monitoring exercise indicate that reliable (i.e high availability) mobile satellite reception is possible in open or limited obstructed conditions, and at vehicle speeds exceeding 120 km/hr, using very modest link margins (of the order of a few dB’s). Clearly, in built-up areas, obstruction of the satellite-to-receiver path is the limiting factor on satellite coverage and, for this reason, the nature of the environment in the vicinity of the receiver and satellite elevation angle are major factors affecting service availability. As the Eureka system is designed to make constructive use of echoes, it is possible to supplement the satellite signal with on-channel terrestrial repeaters to ‘fill’ shadowed areas created by large buildings and other obstructions (the hybrid satellite broadcast concept). In rural (& possibly some suburban) environments, the frequency and depth of signal fades is considerably less and, therefore, inclusion of an appropriate fade allowance in the satellite down-link budget should ensure reliable service in these situations.

It is interesting to note that Australia is more fortunate than most countries with an interest in satellite sound broadcasting. For instance, one can surmise that open reception conditions are characteristic of most rural and remote (in-land) regions of Australia. Also, as a low-to-mid latitude country, the range of elevation angles available from a geostationary orbit (GSO) are reasonably high, so avoiding the need for more complex (and expensive) highly-inclined elliptical orbit (HEO) systems, under consideration by some high latitude countries [3,4]. As an example, Australia’s proposed DBSTAR system [5], at a GSO location of 151.5°E, will provide arrival angles ranging from 35° (Perth, WA) to over 70° (Cooktown, QLD), with well over 90% of locations achieving elevations greater than 40°.

With the multi-carrier modulation scheme (COFDM) used in the Eureka system, non-linearities in the amplification process will produce both in-band and out-of-band intermodulation products. These unwanted products can cause interference to adjacent channels, as well as an increase in the apparent noise of the amplifier.

Some investigations were undertaken into the impact of transponder operating point on the intermodulation levels generated by the non-linearised SSPAs aboard the B3 spacecraft. The down-link COFDM spectrum was monitored at the OPTUS Belrose (Sydney) earth station using a 4.6 metre dish with calibrated L-Band down converter and measurement receiver. With the Eureka system, a measure of unwanted products is the ‘DAB linearity’. This is defined as the level of attenuation of the side lobes, present at the band edges of the COFDM emission, relative to the main signal spectrum. Table 1-2 shows the measured DAB linearity for three different operating points of the L-Band transponder. The corresponding DAB linearity of the up-linked Ku-Band waveform exceeded 30 dB.

⁽¹⁾ Operating point for audio monitoring programme

Current work in the design of L-Band amplifiers for the terrestrial application suggests that at least 30 dB (DAB) linearity is required in order to minimise system degradations [6,7,8]. It is evident from Table 1-2 that this level of intermod attenuation will entail considerable output back-off. Increasing transponder power to compensate may not be feasible given the cost implications, as well as the system constraints on a satellite design (i.e available bus power). In practice, it is likely that some limited back-off, together with other measures, such as appropriate selection of amplifier class for the type of modulation, linearity correction and, possibly, additional processing of the signal to reduce the crest-factor of the waveform, will be necessary to achieve an optimum trade-off between amplifier gain, efficiency and linearity.

There is also need for additional work to quantify degradations, at the receiver, of reduced levels of DAB linearity. The encouraging results from the audio monitoring programme, despite 20 dB linearity and a low signal margin, suggests there is scope for reduction in the nominal 30 dB figure (at least from the effects of in-band products on receiver margin).

The satellite propagation path to a vehicular receiver suffers impairment due to shadowing from surrounding buildings, trees and other obstructions, as well as from frequency selective fading due to reflections from nearby obstacles. To provide acceptable signal availability, a satellite design must therefore include sufficient fade margin in the down-link.

As noted previously, the extent of the fading is largely a function of the environment (i.e open, urban, suburban) in the vicinity of the receiver, as well as the satellite elevation angle. In order to characterise these relationships at L-Band frequencies, a series of signal fade measurements were undertaken in a range of reception environments and at three separate elevation angles - 51 ° , 45.4° and 33.3° . These angles correspond to receive locations in Canberra, Adelaide and Perth respectively, and are representative of the mid to lower range of (geostationary) elevation angles to be encountered in Australia.

8.1 Measurement system

A schematic of the receive measurement system is shown in Figure 1-3. The system includes a Rhode & Schwarz ESVB field strength meter with low noise pre-amplifier, personal computer for data acquisition/storage and a GPS receiver/controller for position fixing. The equipment was installed in a Nissan four wheel drive and powered from a portable petrol generator mounted on the rear loading tray. A peiseler wheel attached to the rear of the vehicle triggered the measurement receiver at regularly spaced intervals.

The receive antenna consisted of a Garmin (GPS) quadrifilar helix with integral low noise amplifier (LNA) mounted on the vehicle roof, at a height above ground of approximately 2.0 metres. The antenna provides near-hemispherical coverage, allowing measurement of not only direct shadow losses but also any fade contributions due to multipath. The measured gain of the antenna/integral LNA was approximately 12 dBic, and the -3 dB elevation angle 10°.

To maximise dynamic range, measurements were based on a carrier-wave (cw) signal, down-linked from the Optus B3 spacecraft at a frequency of 1552.5 MHz. The unfaded carrier-to-noise power measured at the field strength meter varied from 22 to 24 dB, depending on location. By applying an appropriate correction factor (explained in section 8.3) it was possible to measure accurately signal fades of at least 20 dB.

8.2 Measurement procedure

Measurement data was collected over approximately 2200 km of road distance, covering a variety of environments. To assist in categorising reception conditions a terrain and clutter descriptor was applied. The latter was based on the sample land cover classification scheme described in ITU-R Report 567-4. A listing of the various classifications is provided in Table 1-4.

To characterise the combined fade contribution (i.e from shadowing and multipath), measurements of the received signal level were taken at regular intervals of 2.5 cm. Data was recorded, in 'segments’ of 400 metres, along a total of 82 test routes, comprising 21 in Canberra, 30 in Adelaide and 31 in Perth. The length of individual routes varied from 12 to 196 km.

To assist in later processing of the data, a detailed pro-forma log sheet was completed for each test route. The reference (i.e unobstructed line-of-sight) signal, and receiver system noise floor levels were recorded prior to commencement of each measurement run. During the actual run notes were taken, on a segment basis, of changes in the environment type; satellite azimuth bearing with respect to direction of vehicle travel; any major obstructions (eg trees, buildings, terrain, overpasses etc) in the satellite path and, if applicable, (estimated) height and distance from the road. Given the vehicle was travelling at normal road speeds this task was at times demanding, but the information proved useful in the subsequent processing stage - particularly in ensuring only data from homogeneous segments was processed and reconciling any anomalies in the recorded data.

8.3 Data processing

The log sheet(s) were examined in order to validate reception conditions along each test route. Segments not consistent with the terrain/clutter definitions of Table 1-4, or which were identified as invalid (for example, during transitions from one environment type to another) were flagged and omitted from further analysis. Within each run, consecutive segments of the same terrain/clutter type were then concatenated to form a ‘sector’ file. This process generally resulted in several sector files, depending on the number of environment types traversed during the course of each run.

The first phase of processing involved level adjustment of the data to offset any contribution due to receiver system noise. At high carrier-to-noise ratios this offset is negligible, however, at low ratios (i.e near the noise floor of the measurement system) the noise contribution becomes significant and can no longer be ignored. To ensure measurement accuracy was maintained over the range of recorded signal levels, a weighted correction factor was applied to each data point in the file.

In the second phase, cumulative fade distribution functions were generated from the ‘corrected’ files. The data in each sector file was subtracted from the reference line-of-sight measurement and the resulting data file converted to a discrete probability distribution function (PDF) with 0.5 dB bins. The resulting PDF was then integrated to generate the cumulative fade distribution function. The standard deviation of the recorded signal levels in each sector file was also calculated.

In a separate exercise, a moving 41 point (i.e one metre) Hanning window [9,10] was used to average or ‘smooth’ the data set in each corrected file, before a second series of computations to derive the corresponding cumulative fade distribution function. This averaging process serves to remove ‘small’ area effects arising from multipath while preserving the ‘wide’ area component of the signal fade due to shadowing. The relatively short averaging distance ensures that high frequency shadowing due to small obstructions, such as individual trees, is also preserved.

8.4 Presentation of results

The detailed results of the fade measurement programme are presented in Annex 2 & 3 of this report. Annex 2 shows the cumulative fade distribution (CFD) function for the eight environment types (open, rural, wooded, suburban, dense suburban, urban, dense urban and highrise), at each of the three elevation angles (51° , 45.4° and 33.3° ). For each combination of environment type and elevation angle, two graphs have been generated. The ‘narrow-band’ graph is derived directly from the measured fade data, and contains both shadowing and multipath information, while the corresponding ‘smoothed narrow-band’ graph incorporates the additional processing (described in the previous section), to reveal the effect of the shadowing component alone.

For each graph three CFD curves are shown. The inner and outer curves represent the ‘best’ and ‘worst’ sectors (respectively), and indicate the range of fade variation encountered for a given environment type and elevation angle. The third curve is obtained by pooling all sector files, belonging to the particular environment/elevation angle, and then calculating the CFD function. The size of the pooled data base is indicated by the distance (dist) figure. As an example, Figure A2-25 shows the narrow-band (i.e measured cw) CFD curves for wooded environments at 45.4° (Adelaide). For 90% of locations, measured signal fades were less than 2 dB (for the best-case sector) or 11.3 dB (for the worst-case sector). With all data from wooded sectors at this elevation angle pooled (providing a data base extending over 68.8 km of road distance), the signal fade was less than 7 dB for 90% of locations.

A similar analysis can be applied to the smoothed narrow-band graphs (i.e the shadowed component of the cw signal fade), although in this case there is need for some caution. Because of the limited dynamic range of the measurement system, application of the smoothing algorithm results in a steep roll-off in the CFD function, as fade levels approach the noise floor of the system (i.e 22 to 24 dB). At these levels the CFD curves will exhibit less fading (due to shadowing) than is actually the case. Despite this limitation, close examination of the data confirms that the smoothed curves are not affected for fade levels up to at least 15 dB.

Annex 3 contains the tabulated narrow-band fade data (50%, 10% & 1%), on a sector basis, for each elevation angle. Each numbered sector includes the terrain and clutter descriptor, azimuth bearing (of the satellite), path length, and standard deviation of the recorded signal levels along the sector. Those entries which also correspond to a ‘best’ or ‘worst’ case CFD sector in Annex 2 are identified by ‘b’ or ‘w’ (respectively) in the first column. For instance, the best-case sector in the previous example can be identified as sector no. 30 in Table A3-2 of Annex 3. In this particular case, the bearing of the satellite (30° ) is roughly in line with the direction of travel, which explains the low level of fade despite the wooded environment.

9.1 Open areas

In open environments there is evidence of some fading, however this is largely due to uncertainties associated with the measurement system including antenna gain variations with elevation and azimuth and, to a lesser extent, the coarseness of the bin size (0.5 dB) used in generating the PDFs.

9.2 Rural, suburban and dense suburban areas

The rural and suburban (including dense) CFD curves show fades in some sectors approaching 15 to 20 dB (at the 1% level). However, this level of fading is relatively infrequent as illustrated by the pooled CFD curves which show substantially less fade overall. On the basis of the pooled data, close to 95% of rural and suburban locations, at the three elevation angles, would be served with a fade margin of 5 dB. The same fade allowance would ensure just over 90% location availability in the case of dense suburban environments, although this figure does not include measurement data for Perth. In all three environments, tree/foliage obstruction is the dominant cause of signal impairment. This is reflected in the CFD curves, which exhibit very similar fade distributions.

9.3 Wooded areas

The CFD curves for this environment are characterised by a broader distribution and higher ‘knee’ in the function, indicating increased signal attenuation for a greater percentage of locations. In the worst-case sectors, signal fades are seen to exceed 20 dB (1% level) at all three elevation angles. In these same sectors, signal attenuation exceeds 10 dB in about 10 to 25% of locations, depending on elevation. The increased impairment is also reflected in the pooled data. Here, the CFD curves show fades exceeding 10 dB for about 2% (Canberra), 5% (Adelaide) and 10% (Perth) of locations.

9.4 Urban, dense urban and highrise areas

In urban, dense urban and highrise environments, building obstruction/clutter is the dominant cause of signal impairment. Notwithstanding the limited dynamic range of the measurement system, the general trend in the distributions suggest that fade levels approaching 25 to over 30 dB are possible in all three environments. The high knee and shallow roll-off evident in the dense urban and highrise curves (particularly at the lower elevations) also indicates the presence of severe shadowing. In the case of Perth, the worst-case curves show signal fades exceeding 10 dB for 20% (urban), 45% (dense urban) and 70% (highrise) of locations. The corresponding figures based on the pooled data are 7%, 18% and 35% respectively. It is clear that reliable mobile satellite reception in these environments will be very difficult to achieve.

The low degree of fading evident in the Canberra data is due to a number of factors. Apart from the high elevation angle, the city is characterised by relatively wide streets, and few built-up districts. These districts also tend to be less cluttered (in terms of building density) with generally greater separations and lower heights, than encountered in most other cities. In case of the urban curves, the data base consists of only four short sectors of which two were in line with the path to the satellite, also resulting in minimal obstruction. A similar situation exists with the Adelaide urban curves. Here, all three sectors in the data base are offset from the satellite bearing by 30°, resulting in less path obstruction than might otherwise have been the case.

As noted previously, a mobile receiver is subject to fading from both shadowing and multipath, resulting in large variations in the level of the received signal. A measure of signal variability is obtained from the standard deviation (SD) of the received levels. Table 1-5 presents SD values for each environment/elevation angle, as derived from the corresponding pooled data base. Both narrow-band (NB) and smoothed narrow-band (S-NB) values are shown.

The tabulated narrow-band data illustrates the effect of reception environment and elevation angle on overall signal variability. Those environments characterised by high clutter levels (for example, wooded, urban and highrise) exhibit larger SD values, indicating greater variation of the received signal. A reduction in satellite elevation angle also results in larger signal variation, for a given environment type. Although there are individual exceptions, the general trend is clearly evident from the data set.

Table 1-5 also shows the relative contribution of shadowing and multipath to overall signal variability. It is important to bear in mind that the narrow-band SD values reflect overall signal variation (i.e from shadowing and multipath), while the corresponding smoothed narrow-band values reflect only the shadowing component. These results show that the differences in standard deviations are relatively small - ranging from 0.07 dB (open) to 0.78 dB (wooded), which suggests that multipath is not a major factor in the variability of the received signal. In other words, the narrow-band SD’s are more representative of the shadowing component than the multipath component. This appears to be the case across the range of environment types and elevation angles presented.

Another measure of the multipath contribution is obtained by comparing the narrow-band and smoothed narrow-band CFD’s in Annex 2. It is observed that in cluttered environments, the narrow-band curves exhibit more fading at low levels of received power (i.e at the higher fade depths). This usually corresponds to situations where the line-of-sight path is heavily attenuated (i.e shadowed) and the presence of low-level multipath contributes to increased fading due to signal cancellation between the scattered signal components. In this case, multipath is an important factor in the variability of the received signal. At high levels of received power (i.e at the lower fade depths) the line-of-sight path, although somewhat attenuated, is the dominant component and the received signal is much less affected by low-level multipath. Here, shadowing is the major factor and both the narrow-band and smoothed narrow-band curves exhibit similar fade characteristics.

The impact of multipath is quantified in Table 1-6 which shows the relative fade contributions from both shadowing and multipath at 90% location availability. The data is derived from the pooled CFD curves.

The table shows that in less cluttered environments the multipath fade contribution is minimal. Indeed, many of the tabulated values are within the uncertainty associated with the PFD bin size, so it is questionable whether even these marginal contributions are realised.

The multipath is most apparent in wooded, highrise and, to a lesser extent, dense urban environments. In the former case, results show a 1.1 dB contribution at a shadowed fade level of 9.2 dB. Not surprisingly, this is observed at the lowest elevation angle. In highrise environments the multipath contributions are somewhat larger, but the values are realised at shadowed levels exceeding 15 dB. The (relatively) high multipath contribution evident in wooded areas is attributed to extensive ‘close-in’ scattering of the signal from overhanging foliage and road-side trees. Although not tabulated here, the corresponding results for the worst-case sectors - which show most signal variability - lead to similar findings.

This report has described a series of L-Band satellite trials using the Optus B3 spacecraft. A Eureka 147 multiplex signal was successfully transmitted from an L-Band communications payload aboard the satellite, and the reception of a CD quality audio material demonstrated under both mobile and fixed site conditions. These were the first tests of mobile satellite reception using the Eureka system.

The results of an extensive audio monitoring programme, extending over some 4,400 km of road distance, confirm that reliable mobile satellite reception is possible in open or limited obstructed environments, using very modest link margins (of the order of a few dB’s).

In addition, a comprehensive series of L-Band propagation measurements were undertaken using a carrier-wave signal source. The measurements were made in a variety of environments, and at three separate locations, representative of the mid-to-lower range of (geostationary) elevation angles in Australia. Results obtained from this exercise indicate that reliable coverage (i.e exceeding 90%) is possible in open, rural and suburban environments with a fade allowance of around 5 dB. In more cluttered environments (i.e wooded, urban, dense urban and highrise), the depth and extent of the signal fades is much greater, and cannot be compensated without excessive link margins. For these environments, signal diversity (for example, through use of terrestrial repeaters) is the only viable option for significantly improving coverage.

Analysis of the fade data revealed that multipath is a factor only in heavily cluttered environments, and then, at received signal levels some 10 dB or more below the unobstructed line-of-sight signal. At higher levels of received power, shadowing is the dominant component. As satellite systems are power limited, margins of more than 5 to 10 dB, at the receiver, are unlikely. Any significant contribution due to multipath will, effectively, be masked by the noise threshold of the receiver. As a consequence, the major constraint on service availability is shadowing.

The low level of the multipath reflections also means that frequency diversity and/or receiver processing techniques designed to enhance reception in a frequency selective (i.e multipath) environment are of little consequence in a mobile satellite-only channel. This leads to the conclusion that, for the range of elevations considered, narrow-band propagation measurements may also provide a useful measure of the performance of wide-band COFDM systems.

This report follows an extensive series of satellite trials undertaken by the Communications Laboratory, and was made possible through the support of a number of organisations. The authors wish to thank the following for their substantial contribution to this study:

Optus Communications Pty Ltd for providing access to the Optus B3 L-Band transponder, and for provision of up-link facilities during the first phase of the trials.

Optus Belrose Broadcast Operations Centre staff for operational assistance throughout both phases of the trials.

Australian Broadcasting Corporation for loan of the Ku-Band up-converter and HPA, and for technical support in establishing an up-link at the Communications Laboratory for the second phase trials.

National Transmission Agency for providing personnel to assist in the Canberra and Adelaide section of the fade measurement programme.

Special Broadcasting Service for supporting initial proposals to undertake the satellite trials and for coordinating the technical assistance provided by the Australian Broadcasting Corporation.

Communications Laboratory staff involved directly, and in support, of the extensive measurement programme, as well as the subsequent data processing and analysis stage.

Plate 1: Preliminary testing of the receiver low noise amplifier/down converter constructed for the satellite trials. Note the bifilar helix rod and planar directional antenna at rear-left of photograph.

Plate 2: Close up view of the receiver low noise amplifier/down converter.

Plate 3: Nissan 4WD used for mobile satellite measurements, near Albany (Western Australia). Note peiseler wheel in stowed position. Photograph also shows bifilar helix rod antenna (for reception of the Eureka DSB signal) and Garmin omnidirectional antenna (for fade measurements).

Plate 4: View of satellite reception/measurement equipment rack in Nissan 4WD. Note receiver low noise amplifier/down converter, Eureka 147 third generation receiver and ESVB field strength meter (directly above receiver).

Plate 5: View of the 4.5 m earth station antenna (Belrose, Sydney ) used for up-linking during the Canberra phase of the satellite trials.

Plate 6: View of the up-converter & klystron amplifier at the Belrose earth station.

Plate 7: View of the 4.6 m earth station antenna (Communications Laboratory, Canberra) used for up-linking during the Adelaide and Perth phase of the satellite trials. Note: shelter at rear used for housing up-link equipment.

Plate 8: Inside view of shelter showing DSB equipment. Philips encoder visible on top-left of rack.

Plate 9: Inside view of shelter showing Ku-Band up-converter, TWT amplifier and spectrum analyser (for local monitoring of DSB signal ). Signal Generator (at top of rack) provided the cw signal for the fade measurements.

Plate 10: Photograph shows video camera mounted on Nissan 4WD measurement vehicle. Approximately 80 minutes of audio/video recordings were made in and around Canberra to demonstrate audio quality in a variety of reception conditions.